- Date: 5 Jul. (Thu.)
- Time: 16:30 — 18:00
- Place: W.W. 7th-floor
- Speaker: Jun Sekine
- Title: Some remarks on Bayesian optimal CRRA-utility
- Abstract:
A Bayesian CRRA-utility maximization is considered, where mean-return-rates of risky assets
is an unobservable random vector.
Several remarks are mentioned for this problem:
Several remarks are mentioned for this problem:
i) The optimal expected utility grows hyperbolically in the long run when Arrow-Pratt’s risk aversion parameter is bigger than 1. This provides a sharp contrast to the results of “non-Bayesian” settings.
ii) For log utility case with infinite horizon, where optimal portfolios are sometimes called Bayesian Kelly portfolio, or universal portfolio, reduction techniques of computational difficulties are introduced: EG-update portfolio, and Gaussian prior combined with projection will be mentioned.
The first part is based on a joint work with Hideaki Miyata of Kyoto University, and the second part is an ongoing joint research with Akifumi Isogai and Yusuke Tashiro of Mitsubishi UFJ Trust Investment Technology Institute.
日本語
English